OLAP انجمن حکومت کاوی مبتنی بر سیستم مدیریت کیفیت ترکیبی برای استخراج الگوهای نقص در صنعت پوشاک
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|4490||2013||12 صفحه PDF||سفارش دهید||7320 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Expert Systems with Applications, Volume 40, Issue 7, 1 June 2013, Pages 2435–2446
In today’s garment industry, garment defects have to be minimized so as to fulfill the expectations of demanding customers who seek products of high quality but low cost. However, without any data mining tools to manage massive data related to quality, it is difficult to investigate the hidden patterns among defects which are important information for improving the quality of garments. This paper presents a hybrid OLAP-association rule mining based quality management system (HQMS) to extract defect patterns in the garment industry. The mined results indicate the relationship between defects which serves as a reference for defect prediction, root cause identification and the formulation of proactive measures for quality improvement. Because real-time access to desirable information is crucial for survival under the severe competition, the system is equipped with Online Analytical Processing (OLAP) features so that manufacturers are able to explore the required data in a timely manner. The integration of OLAP and association rule mining allows data mining to be applied on a multidimensional basis. A pilot run of the HQMS is undertaken in a garment manufacturing company to demonstrate how OLAP and association rule mining are effective in discovering patterns among product defects. The results indicate that the HQMS contributes significantly to the formulation of quality improvement in the industry.
Nowadays, customers are seeking products of high quality and low cost. Manufacturers are urged to achieve better quality in their products so as to stay competitive in the industry. Unfortunately, variance in product quality is unavoidable as it can be induced by many factors during production. One of these critical factors is the workmanship which commonly exists in labor-intensive industries where many production processes are performed manually. In the garment industry, human factors such as different levels of skill, years of experience and human errors may result in garment defects in some circumstances. In order to produce high-quality and low-cost products, it is important to achieve quality improvement while at the same time to identify any product defects at an early stage. Unfortunately, it is challenging to maintain the quality of garments which are processed manually. It is thus necessary to inspect products carefully so as to ensure they are of good quality. Traditionally, garment defects are identified by human inspectors who treat each defect individually without being aware of the relationship between different defects, thus making causal analysis and defect prediction difficult. Fig. 1 depicts the existing problems in handling quality problems in the garment industry. There are various departments responsible for different tasks along the production workflow from product design to final product inspection. Hence, it is difficult for manufacturers to identify the department to which a particular garment defect should be attributed, and the root causes of the defects. This shows that there is a lack of information to tackle product quality problems. Manufacturers do not have timely information to analyze defect causes and individual departments fail to be aware of any possible defects that they might be causing. In addition, owing to the complexity of garment manufacturing processes, there are numerous defects which can be found on a single garment. Without any tools to manage massive relevant data and identify the hidden pattern among defects, manufacturers are unable to discover any correlations between defects, or the reasons for different defects. This indicates the lack of a mechanism for investigating defect patterns which could be useful in defect prediction and defect diagnosis. The problems outlined above will certainly lead to bad consequences for the garment industry, such as failure in achieving quality improvement, low customer satisfaction, high rework cost and long production cycle times. With the aim of tackling these problems, this paper presents an intelligent system, namely hybrid OLAP-association rule mining based quality management system (HQMS), to extract garment defect patterns in the form of association rules so as to formulate useful quality improvement plans. To provide manufacturers with the ability to explore different kinds of desired data effectively and mine data at different levels, Online Analytical Processing (OLAP) is also applied in the system that is developed.This paper is organized as follows: Section 2 is a literature review related to this study. In Section 3, the architecture of the HQMS is proposed. Section 4 contains a case study where a pilot run of the system is conducted in a case company. In Section 5, the discussion of the HQMS is presented. Finally, Section 6 is the conclusion.
نتیجه گیری انگلیسی
This paper presents an intelligent system for quality improvement with the integration of data warehousing, OLAP and association rule mining for extracting garment defect patterns. In the usual practice in the garment industry, individual garment defects are solely identified by human inspection without any references to the correlation between defects. Hence, it is difficult to predict defects and take proactive measures for quality improvement. In the HQMS, the data warehouse is used to manage and store the data for data mining while the connection between OLAP and the association rule mining model allows data mining to be applied on a multidimensional basis so that users can gain more in-depth knowledge for defect diagnosis. The garment industry is considered a good example of a labor-intensive industry where product quality is affected by workmanship, making it difficult to predict product defects. Defects which commonly happen concurrently are believed to have certain hidden correlations. Through the implementation of the HQMS, the results show that the proposed methodology effectively extracts hidden relationships among defects based on their co-occurrence, which in turn tell managers which defects are most likely to occur. This allows easy identification of significant root causes of defects, prediction of their occurrence and provides knowledge support in the formulation of effective quality improvement plans. This research study makes a significant contribution by making use of quality-related data to predict product defects and to perform causal analysis in the garment industry. Although every industry has its unique product characteristics, the HQMS’ structure presented in this paper, besides being of use in the garment industry, is also applicable to most manufacturing industries, in particular, to some labor-intensive industries where product quality is extremely difficult to maintain. Future research might include real-time capturing of product quality data upon inspection, in order to enhance the efficiency of the system